The use of yeast artificial chromosomes (YACs) as vectors for cloning and analysis of the human genome is seriously hampered by high frequencies of cloning gaps and unstable YACs, Recombination is often cited as the cause of YAC instability. However, we propose that the recombination is initiated by DNA breaks that occur as a consequence of defects in chromosome replication or segregation. The presence or absence of sequences in human inserts that fortuitously mimic functional yeast chromosomal elements would contribute to these defects. In particular, GC-rich regions of the human genome that are abundant in genes are expected to be deficient in sequences that can be used as replication origins in yeast Incomplete replication of chromosomes at the time of mitosis will generate broken molecules. This problem will be exacerbated by the fortuitous presence of replication fork barriers or elements that delay origin activation until late in the 5-phase, Sequences fortuitously recognized as centromeres by yeast will create dicentric YACs that also break at mitosis. Breakage events lead to deletions and rearrangements that are eventually stabilized by recombination. Eliminating recombination will not eliminate the basic problem. Identifying and eliminating the cause of the breaks are essential. We propose to develop simple plasmid assays of human genomic DNA, using functional tests in yeast, to detect origin deficiency as well as the presence of fortuitous centromeres, replication fork barriers and late initiation determinants. These assays will be used to assess the contribution of these impediments to stable human YAC cloning by (i) Screening human X or cosmid clones to determine the frequency of each type of YAC cloning barrier in the human genome. We will use random clones and those from gene-rich GC isochores. (2) Testing for these different YAC cloning barriers in specific regions of the human genome that have been unclonable as stable YACS but which exist as lambda or cosmid clones. We propose to then test approaches for alleviating the YAC instability problems posed by the absence or presence of these fortuitous elements. The centromere problem is easily solved by making a second YAC library using vector arms lacking a centromere. Most of the other problems can be alleviated by using yeast strains that have a reduced sequence specificity for origin recognition, thereby creating more origins in the human insert. Many candidate yeast strains already exist.